cellular heterogeneity during mouse pancreatic …gel bead in emulsion (gem) was incubated and then...
TRANSCRIPT
1
Cellular heterogeneity during mouse pancreatic ductal adenocarcinoma progression at 1
single-cell resolution 2
3
Abdel Nasser Hosein1,3,7, Huocong Huang1, Zhaoning Wang2, Kamalpreet Parmar5, Wenting 4
Du1, Jonathan Huang7, Anirban Maitra6,7, Eric Olson2, Udit Verma3, Rolf A. Brekken1,4 5
6
1Hamon Center for Therapeutic Oncology Research, Division of Surgical Oncology, Department 7
of Surgery, 2Department of Molecular Biology, 3Department of Internal Medicine – Division of 8
Hematology & Oncology, 4Department of Pharmacology, and 5Department of Pathology, 9
University of Texas Southwestern Medical Center, Dallas, Texas, USA. 6Department of 10
Pathology and 7Department of Translational Molecular Pathology, University of Texas MD 11
Anderson Cancer Center, Houston, TX, USA. 12
13
Short title: scRNA-seq of PDA GEMMs 14
15
Corresponding author 16
Rolf A. Brekken, PhD 17
Hamon Center for Therapeutic Oncology Research 18
University of Texas Southwestern Medical Center 19
6000 Harry Hines Blvd. 20
Dallas, TX 75390-8593 21
Tel: 214.648.5151; Fax: 214.648.4940 22
24
Disclosures 25
The authors have no conflicts of interest to report. 26
27
Writing assistance 28
Editorial assistance was provided by Dave Primm of the UT Southwestern Department of 29
Surgery. 30
31
Funding 32
This work was supported by NIH grants R01 CA192381 and U54 CA210181 Project 2 to RAB, 33
the Effie Marie Cain Fellowship to RAB, the H. Ray and Paula Calvert Pancreatic Cancer 34
Research Fund to UV, NIH grants U01 CA200468, U01 CA196403, R01 CA218004, R01 35
CA204969, P01 CA117696 to AM and grants from the NIH and Welch Foundation to EO. 36
All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.. https://doi.org/10.1101/539874doi: bioRxiv preprint
2
Author contributions 37
38
ANH: study concept and design, acquisition of data, analysis and interpretation of data, drafting 39
of the manuscript. HH: study concept and design, acquisition of data, analysis and interpretation 40
of data, drafting of the manuscript. ZW: bioinformatics analyses, interpretation of data. KP: 41
pathologic interpretation and scoring of human tissue samples. WD: acquisition of data. JH: 42
bioinformatics analyses, critical review of the manuscript. AM: bioinformatics analyses, 43
interpretation of data, critical review of the manuscript. EO: bioinformatics analyses, 44
interpretation of data, critical revision of the manuscript. UV: study concept and design, critical 45
revision of the manuscript. RAB: study concept and design, interpretation of data, drafting of 46
the manuscript. 47
48
Abbreviations 49
ADM, acinar-to-ductal metaplasia 50
BET, bromodomain and extraterminal 51
CAF, cancer-associated fibroblast 52
FB, fibroblast population 53
GEM, gel bead in emulsion 54
GEMM, genetically engineered mouse model 55
GO, gene ontology 56
PDA, pancreatic ductal adenocarcinoma 57
PSC, pancreatic stellate cell 58
scRNA-seq, single-cell RNA sequencing 59
UMI, unique molecular identifiers 60
61
All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.. https://doi.org/10.1101/539874doi: bioRxiv preprint
3
Abstract: 62
Background & Aims 63
Pancreatic ductal adenocarcinoma (PDA) is a major cause of cancer-related death with limited 64
therapeutic options available. This highlights the need for improved understanding of the biology 65
of PDA progression. The progression of PDA is a highly complex and dynamic process featuring 66
changes in cancer cells and stromal cells; however, a comprehensive characterization of PDA 67
cancer cell and stromal cell heterogeneity during disease progression is lacking. In this study, 68
we aimed to profile cell populations and understand their phenotypic changes during PDA 69
progression. 70
71
Methods 72
We employed single-cell RNA sequencing technology to agnostically profile cell heterogeneity 73
during different stages of PDA progression in genetically engineered mouse models. 74
75
Results 76
Our data indicate that an epithelial-to-mesenchymal transition of cancer cells accompanies 77
tumor progression. We also found distinct populations of macrophages with increasing 78
inflammatory features during PDA progression. In addition, we noted the existence of three 79
distinct molecular subtypes of fibroblasts in the normal mouse pancreas, which ultimately gave 80
rise to two distinct populations of fibroblasts in advanced PDA, supporting recent reports on 81
intratumoral fibroblast heterogeneity. Our data also suggest that cancer cells and fibroblasts are 82
dynamically regulated by epigenetic mechanisms. 83
84
Conclusion 85
This study systematically outlines the landscape of cellular heterogeneity during the progression 86
of PDA. It strongly improves our understanding of the PDA biology and has the potential to aid 87
in the development of therapeutic strategies against specific cell populations of the disease. 88
89
Key words: single-cell RNA sequencing; pancreatic cancer; cellular heterogeneity; fibroblasts; 90
macrophages 91
92
All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.. https://doi.org/10.1101/539874doi: bioRxiv preprint
4
Introduction 93
Pancreatic ductal adenocarcinoma (PDA) carries the highest mortality rate of all major 94
malignancies in industrialized countries, with a 5-year survival of 8.5%. Patients are faced with 95
limited treatment options that achieve poor durable response rates, highlighting the need for an 96
improved understanding of PDA disease biology [1]. PDA progression is a complex and 97
dynamic process that requires interaction between cancer cells and stromal cells [2]. It is 98
characterized by the formation of a unique microenvironment consisting of heterogeneous 99
stromal cell populations that include fibroblasts, macrophages, lymphocytes, and endothelial 100
cells. These stromal compartments are critical in driving PDA biology [3]. 101
102
The dynamic phenotypic changes in different cell populations during PDA progression is not 103
fully understood. Gene expression profiling of bulk tissues provides a limited picture of the 104
cellular complexity of the heterogeneous cell populations in PDA. In contrast, single-cell RNA 105
sequencing (scRNA-seq) has the potential to enable gene expression profiling at the level of the 106
individual cell [4] and provides a powerful tool to understand the cellular heterogeneity of PDA. 107
We applied scRNA-seq to investigate gene expression changes of cancer cells and stromal 108
cells during PDA progression in genetically engineered mouse models (GEMMs). This unbiased 109
approach provided evidence of considerable intratumoral cellular heterogeneity, including 110
molecular insights into epithelial and mesenchymal populations of cancer cells and distinct 111
molecular subtypes of macrophages and cancer-associated fibroblasts (CAFs). 112
113
Methods 114
Animal studies 115
KIC, KPC and KPfC mice were generated as previously described [5-7]. Mice were sacrificed 116
when they were moribund: 60 days old for the KIC (n = 3, late PDA) and KPfC (n = 1) or 6 117
months old for the KPC (n = 1). The 2 KIC mice were sacrificed at 40 days old (early PDA) and 118
“normal pancreas” mice (n = 2) were sacrificed at 60 days old. In experiments using more than 119
one mouse, tissues were pooled prior to enzymatic digestion. The KPfC mouse had a pure 120
C57BL/6 genetic background and all others had a mixed background (C57BL/6 with FVB). 121
Ultrasound imaging was carried out under general anesthesia with isoflurane. Mice were 122
euthanized by cervical dislocation under anesthesia. AVMA Guidelines for the Euthanasia of 123
Animals were strictly followed. Tissues were either fixed in 10% formalin for 124
immunohistochemistry or enzymatically digested for single-cell analysis. 125
126
All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.. https://doi.org/10.1101/539874doi: bioRxiv preprint
5
Tissue digestion 127
A 10x digestion buffer was prepared in PBS: collagenase type I (450 units/ml, Worthington 128
Biochemical, Lakewood, NJ), collagenase type II (150 units/ml, Worthington), collagenase type 129
III (450 units/ml, Worthington), collagenase type IV (450 units/ml, Gibco/Thermo Fisher, 130
Waltham, MA), elastase (0.8 units/ml, Worthington), hyaluronidase (300 units/ml, Sigma-Aldrich, 131
St. Louis, MO), and DNase type I (250 units/ml, Sigma-Aldrich). Tumors and pancreas were 132
enzymatically digested into a single-cell suspension. Briefly, freshly dissected tissue was placed 133
into a 10-cm tissue culture dish and a sterile razor blade was used to cut the tissue into fine 134
pieces. Samples were resuspended in PBS and washed twice by centrifuge at 2000 rpm for 3 135
minutes and added to a 50 ml tube containing 1x digestion buffer containing 1% FBS. The tube 136
was incubated on a shaker at 37°C for 60 minutes. Then 35 ml of PBS was added and cells 137
were washed three times prior to filtering out debris using a 70 μm mesh filter. Single cells were 138
resuspended in 100 μl of PBS in preparation for single-cell library creation. Cell viability was 139
measured by trypan blue. Viability was 80% for the normal pancreas and late KIC samples, 75% 140
for the early KIC and KPfC, and 90% for the KPC. 141
142
Single-cell cDNA library preparation and sequencing 143
Library generation was performed using the 10x Chromium System (10X Genomics Inc., 144
Pleasanton, CA). Single-cell suspensions were washed in 1x PBS (calcium- and magnesium-145
free) containing 0.04% weight/volume bovine serum albumin (400 μg/ml) and brought to a 146
concentration of 200-700 cells/μl. The appropriate volume of cells was loaded with Single Cell 3’ 147
gel beads into a Single Cell A Chip and run on the Chromium Controller. Gel bead in emulsion 148
(GEM) was incubated and then broken. Silane magnetic beads were used to clean up the GEM 149
reaction mixture. Read 1 primer sequence was added during incubation and full-length, 150
barcoded cDNA was amplified by PCR after cleanup. Sample size was checked on an Agilent 151
Tapestation 4200 (Agilent, Santa Clara, CA) using DNAHS 5000 tape and concentration 152
determined by a Qubit 4 Fluorometer (Thermo Fisher) using the DNA HS assay. Samples were 153
enzymatically fragmented and underwent size selection before proceeding to library 154
construction. During library preparation, Read 2 primer sequence, sample index, and both 155
Illumina adapter sequences were added. Samples were cleaned up using AMPure XP beads 156
(Beckman Coulter, Brea, CA) and post-library preparation quality control was performed using 157
DNA 1000 tape on the Agilent Tapestation 4200. The final concentration was ascertained using 158
the Qubit 4 Fluorometer DNA HS assay. Samples were loaded at 1.5 pM and run on the 159
All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.. https://doi.org/10.1101/539874doi: bioRxiv preprint
6
Illumina NextSeq500 High Output Flowcell (Illumina, San Diego, CA) using V2.5 chemistry. The 160
run configuration was 26 x 98 x 8. 161
162
Bioinformatic analyses 163
We used Cell Ranger version 1.3.1 (10x Genomics) to process raw sequencing data and the R-164
package Seurat version 2.0 [8] for downstream analyses. Cell clusters were identified via the 165
FindClusters function using a resolution of 0.6 for all samples, using a graph-based clustering 166
algorithm implemented in Seurat. Marker genes for each cluster were computed, and 167
expression levels of several known marker genes were examined. Different clusters expressing 168
known marker genes for a given cell type were selected and combined as one for each cell 169
type. Gene ontology and pathway analysis were performed using the DAVID bioinformatics 170
suite, version 6.8 [9]. 171
172
Histological analysis 173
Formalin-fixed tissues were embedded in paraffin and cut in 5 μm sections. Sections were 174
evaluated by H&E and immunohistochemical analysis using antibodies specific for vimentin 175
(5741, Cell Signaling Technology, Danvers, MA), BRD4 (AB128874, Abcam, Cambridge, MA), 176
Sox9 (AB5535, EMD Millipore, Burlington, MA), CDH11 (NBP2-15661, Novus Biologicals, 177
Centennial, CO), and H3K27ac (AB4729, Abcam). Following an initial antigen retrieval with Tris-178
EDTA-glycerol (10%) buffer and inhibition of endogenous peroxidase activity, the slides were 179
incubated with primary antibody overnight at 4°C. Slides were then incubated with horseradish 180
peroxidase or alkaline phosphatase conjugated secondary antibody (Vector Laboratories, 181
Beringame, CA) for 1 hour at 25°C. This was followed by development using the appropriate 182
chromogenic substrate: DAB, Warp Red or Ferangi Blue (Biocare Medical, Pacheco, CA). In the 183
case of multichannel immunohistochemistry, slides were subsequently stripped using a sodium 184
citrate buffer and by boiling at 110°C for 3 minutes. The procedure was then repeated as above 185
using a different-colored chromogen for development. All human PDA samples were provided 186
by the UT Southwestern Tissue Management Shared Resource and their use was approved by 187
the UT Southwestern institutional review board for the purpose of research. All patient samples 188
were de-identified and interpreted by a board-certified pathologist (KP). 189
190
Results 191
Cellular heterogeneity during PDA progression 192
All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.. https://doi.org/10.1101/539874doi: bioRxiv preprint
7
We sought to determine the composition of single pancreatic cancer cells during progression in 193
GEMMs. Normal mouse pancreas, 40-day-old KIC (KrasLSL-G12D; Cdkn2aflox/flox; Ptf1aCre/+) mouse 194
pancreas, termed “early KIC” (with early neoplastic changes confirmed by ultrasound; 195
Supplementary Fig. 1), and 60-day-old KIC pancreas, termed “late KIC” (Fig. 1A) were freshly 196
isolated and enzymatically digested followed by single-cell cDNA library generation using the 197
10x Genomics platform [10]. Libraries were subsequently sequenced at a depth of more than 198
105 reads per cell. We performed stringent filtering, normalization, and graph-based clustering, 199
which identified distinct cell populations in the normal pancreas and each stage of PDA. 200
201
In the normal mouse pancreas, 2354 cells were sequenced and classified into appropriate cell 202
types based on the gene expression of known markers: acinar cells, islet cells, macrophages, T 203
cells, and B cells, as well as three distinct populations of fibroblasts. Fibroblasts-1, fibroblasts-2, 204
and fibroblasts-3 (Fig. 1B and E) were noted. In the early KIC pancreas (3524 cells sequenced), 205
the emergence of a cancer cell population was observed (9.9% of cells), expressing known PDA 206
markers such as Krt18 and Sox9 [11] (Fig. 1C and F). The acinar cell population was 207
substantially reduced, while there was a marked increase in total macrophages and fibroblasts. 208
Of note, the same three populations of fibroblasts seen in the normal pancreas were identified in 209
the early KIC lesion. Additionally, endothelial cells were observed at this stage. This indicates 210
that the expansion of fibroblasts and macrophages is an early event during PDA development, 211
accompanying tumor initiation. We next characterized the late KIC pancreas (804 cells 212
sequenced) and noted the absence of normal exocrine (acinar) and endocrine (islet) cells (Fig. 213
1D and G). Instead, two distinct populations of cancer cells were present, suggesting 214
phenotypic cancer cell heterogeneity as a late event in the course of the disease. We also 215
observed the presence of only two distinct fibroblast populations, which had a similar 216
percentage in relation to total cells. Noticeably, macrophages became a predominant cell 217
population in the late KIC tumor. Moreover, we observed lymphocytes at this stage. The cellular 218
heterogeneity in cancer cells and stromal cells in the early and late KIC lesions highlighted the 219
dynamic cellular changes that occur during PDA progression. 220
221
Mesenchymal cancer cells emerge in advanced PDA 222
Gene expression analysis of cancer cell epithelial markers (Cdh1, Epcam, Gjb1, and Cldn3) and 223
mesenchymal markers (Cdh2, Cd44, Axl, Vim, and S100a4) revealed that early KIC cancer cell 224
populations assumed an epithelial expression profile (Fig. 2A and C). This is in contrast to tumor 225
cell populations in the late KIC tumors, where we identified two distinct cancer cell populations: 226
All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.. https://doi.org/10.1101/539874doi: bioRxiv preprint
8
one enriched for epithelial markers and the other, more abundant population, enriched for 227
mesenchymal markers (Fig. 2B and C). These data support that tumor cell epithelial plasticity 228
contributes to cancer cell heterogeneity during the progression of KIC tumors. 229
230
The hierarchical clustering of the top significant genes in each of the three cancer cell 231
populations (epithelial cancer cells in early KIC, epithelial and mesenchymal cancer cell 232
populations in late KIC) was performed (Fig. 2D). In addition, gene clusters from the cancer cell 233
populations were subjected to pathway and gene ontology (GO) analysis. First, we compared 234
cancer cells of the early KIC population to the total cancer cells of the late KIC and found that 235
the most downregulated genes in late KIC cancer cells were associated with normal pancreatic 236
function such as pancreatic secretion, digestion and absorption, and insulin secretion (Fig. 2E 237
and F). Moreover, normal pancreatic acinar genes such as Try4, Try5, Cela2a, Cela3b, Reg2, 238
and Rnase1 were expressed at higher levels in early KIC cancer cells, while late KIC cancer 239
cells expressed a higher level of the pancreatic ductal gene Muc1 (Fig. 2D). This is suggestive 240
of an ongoing acinar-to-ductal metaplasia (ADM) during tumor progression in this GEMM. In 241
contrast, the most upregulated genes in late KIC cancer cells were associated with ribosome, 242
glycolysis/gluconeogenesis, and amino acid biosynthesis, which is highly suggestive of 243
increased translation and metabolically active cancer cells in established KIC tumors. 244
Interestingly, pathways previously reported to be closely associated with the stroma and 245
progression of PDA were also highlighted, such as ECM-receptor interaction [12], TGFβ [13], 246
and hippo signaling pathways [14]. We then compared early KIC cancer cells with the late KIC 247
epithelial cancer cell population to understand the mechanisms that promoted the progression 248
of PDA in the epithelial cancer cell compartment. Interestingly, similar cell functions/signaling 249
pathways were identified by comparing the two epithelial cancer cell populations (Fig. 2G and 250
H). Taken together, these analyses objectively demonstrate an ADM state during the 251
progression of KIC tumors and suggest that stroma-cancer cell interaction promotes the 252
progression of PDA and cancer cell heterogeneity. 253
254
Mesenchymal cancer cells exist in advanced PDA GEMMs with different diverse mutations 255
In addition to KRAS mutations, additional driver events are required for PDA progression [8], 256
with TP53 and INK4A being the second- and third-most commonly mutated genes in human 257
PDA, respectively. As such, we sought to understand the effect of different secondary driver 258
mutations on the phenotypes and heterogeneity of cancer cells. We performed scRNA-seq in 259
another PDA GEMM, KPfC (KrasLSL-G12D; Trp53Flox/Flox; Pdx1Cre/+) (Fig. 3A). Consistent with late 260
All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.. https://doi.org/10.1101/539874doi: bioRxiv preprint
9
KIC tumors, two distinct cancer cell populations expressing Krt18 and Sox9 were noted in late 261
KPfC (60-day-old) tumors (Fig. 3A and B), one marked by epithelial markers such as Gjb1, Tjb1, 262
Ocln, and Cldn3, while the other was marked by mesenchymal markers such as Vim, Cd44, Axl, 263
S100a4, and Fbln2 (Fig. 3C and D). Epithelial and mesenchymal cancer cell populations in 264
KPfC mice shared many genes in common with the corresponding populations in KIC; however, 265
they also expressed unique gene signatures (Fig. 3F). 266
267
We then compared the total cancer cell gene signatures between late KIC and late KPfC mice 268
by KEGG and Biocarta pathway analysis methods, in an attempt to identify potential differences 269
in cancer cell signaling pathways caused by the different secondary driver mutations. As 270
expected, the p53 signaling pathway was upregulated in the KIC model by comparison to the 271
KPfC model (Fig. 3E). The analyses of late KIC and late KPfC mice suggests that cancer cell 272
heterogeneity is a late-stage tumor event that occurs in the setting of multiple secondary driver 273
mutations. However, under the same oncogenic Kras mutation, different secondary driver 274
mutations can potentially lead to different signaling pathways that drive PDA progression. 275
276
Macrophage heterogeneity during PDA progression 277
We found a marked increase in the size of the macrophage population as PDA progressed from 278
normal pancreas to early KIC and eventually late KIC tumors (Fig. 1B-D). We further 279
characterized the macrophage compartment during PDA progression by subclustering 280
macrophages in early and late KIC tumors, which revealed three transcriptionally distinct 281
macrophage clusters in early KIC and two in late KIC (Fig. 4A and C). 282
283
Macrophage population 1 in early KIC tumors was characterized by the expression of Fn1, 284
Lyz1, Lyz2, Ear1, and Ear2 as well as Cd14 (Fig. 4B). Moreover, these macrophages 285
specifically expressed high levels of the IL1 receptor ligands: Il1a, Il1b, and Il1rn. GO analysis 286
suggested that this macrophage population was involved in healing during inflammation, the 287
regulation of type I and III hypersensitivities, and antigen processing and presentation (Fig. 4E). 288
In contrast, macrophage population 2 was noted to express an abundance of chemokines, 289
including Ccl2, Ccl4, Ccl7, Ccl8, and Ccl12, as well as many complement-associated genes 290
(Fig. 4B). Indeed, leukocyte activation, complement activation, and humoral response genes 291
were the most significantly enriched GO categories in this macrophage population (Fig. 4E). 292
The third macrophage population expressed Ccl17 and Ccr7 and was enriched in ribosomal 293
small-unit biogenesis, translation, and antigen-processing functions (Fig. 4B and E). Importantly, 294
All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.. https://doi.org/10.1101/539874doi: bioRxiv preprint
10
macrophages in normal mouse pancreas weakly expressed genes found in macrophage 295
population 2 and 3 from early KIC mice, suggesting that the normal pancreas macrophages 296
could be noncommitted macrophages residing in tissue in the normal organ that are induced to 297
adopt a distinct phenotype upon tumor initiation (Fig. 4B). 298
299
The late KIC tumor featured two macrophage subpopulations (Fig. 4C). Macrophage population 300
1 highly expressed genes such as S100a8 and Saa3, which have been shown to be expressed 301
in lipopolysaccharide-treated monocytes [15]. Moreover, numerous chemokines were elevated 302
in this population such as Ccl2, Ccl7, Ccl9, Ccl6, Cxcl3, and Pf4 (Fig. 4D). GO analysis revealed 303
this population is likely associated with Stat3 activation, leukocyte chemotaxis, and response to 304
lipopolysaccharide and inflammatory stimuli (Fig. 4F). These data suggest that macrophage 305
population 1 was inflammatory in nature. Macrophage population 2 of late KIC tumors was rich 306
in MHC-II antigen presentation molecules: Cd74, H2-Aa, H1-Ab1, H2-Dma, H2-Dmb1, H2-307
Dmb2, and H2-Eb1 (Fig. 4D), and GO analysis highlighted antigen presentation and adaptive 308
immune response pathways as being elevated (Fig. 4F). Consistently, in late KPfC tumors, we 309
also observed two distinct populations of macrophages with similar features (Supplementary 310
Fig. 3). Interestingly, we did not observe a macrophage population in late tumors that correlated 311
with macrophage population 1 from the early tumors, suggesting that this population might 312
undergo negative selection or a differentiation into inflammatory and/or MHC-II–rich 313
macrophages during tumor progression. 314
315
We also compared the features of the total macrophage clusters between early and late KIC 316
tumors and observed a substantially enhanced macrophage inflammatory signature as the 317
tumor progressed (Fig. 4G). A wide variety of inflammatory genes increased, including Il1a, Il1b, 318
Il1r2, and Il6. GO analysis of this gene list highlighted leukocyte chemotaxis and inflammatory 319
response functions as increased in advanced KIC tumors (Fig. 4H). These data suggest that 320
PDA progression is characterized by an increase in inflammatory features in macrophages. 321
322
Fibroblast heterogeneity during PDA progression 323
In normal pancreas and early KIC tumors, we had identified three distinct populations of 324
fibroblasts, while in late KIC only two fibroblast populations were noted (Fig. 1B-D). To ascertain 325
the relationship between these fibroblast populations and the dynamics of their phenotypic 326
changes during PDA progression, we projected fibroblasts from the three analyses onto a single 327
tSNE plot and applied a graph-based clustering algorithm (Fig. 5A) which revealed three distinct 328
All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.. https://doi.org/10.1101/539874doi: bioRxiv preprint
11
molecular subtypes of fibroblasts in the normal pancreas, early KIC tumors, and late KIC 329
tumors. The overlay demonstrates that the normal pancreas and early KIC tumors contained all 330
three fibroblast subtypes while the late KIC contained only two (Fig. 5A), confirming our initial 331
analysis (Fig. 1B-D). Specifically, this analysis demonstrated that fibroblast population 1 (FB1) 332
and fibroblast population 3 (FB3) found in normal and early KIC pancreas were present in the 333
late KIC tumor whereas fibroblast population 2 (FB2) was absent. 334
335
In the normal pancreas, FB1, FB2, and FB3 made up 35.4%, 56.9% and 7.7% of the total 336
fibroblasts, respectively (Supplementary Fig.4A). In early KIC tumors, although the total 337
fibroblasts expanded (Fig. 1C), the ratios of each fibroblast population remained similar. 338
Furthermore, in the late KIC tumors, FB1 and FB3 were present in nearly equal proportions of 339
46.5% and 53.5%, respectively (Supplementary Fig.4A). Each fibroblast population was 340
characterized by distinct marker genes. For example, FB1 markedly expressed Cxcl14, Ptn, and 341
several genes mediating insulin-like growth factor signaling such as Igf1, Igfbp7, and Igfbp4. 342
FB2 specifically expressed Nov, a member of the CCN family of secreted matricellular proteins 343
[16] as well as Pi16, which has been shown to be expressed in fibroblast populations in various 344
tissue types [17], in addition to Ly6a and Ly6c1. FB3 showed distinct expression of mesothelial 345
markers such as Lrrn4, Gpm6a, Nkain4, Lgals7, and Msln [18] in addition to other genes 346
previously shown to be expressed in fibroblasts such as Cav1, Cdh11, and Gas6 [19-21]. 347
348
Hierarchical clustering of the most significant genes for each fibroblast subtype confirmed the 349
persistence of FB1 and FB3 during the progression of PDA (Fig. 5B) and that they exist across 350
different advanced-stage PDA GEMMs (KPC and KPfC), suggesting a consistent cell of origin. 351
Interestingly, the gene expression heatmap also indicated that the FB2 population started to 352
move toward an FB1-like expression profile in early KIC tumors, suggesting FB1 and FB2 might 353
converge into a single CAF population with FB1 features by late invasive disease. Of note, Il6, 354
Ccl2, Ccl7, Cxcl12, and Pdgfra were expressed in FB1 and FB2 in the normal pancreas and 355
early KIC tumors, and showed greater expression in FB1 of late KIC (Fig. 5C). In contrast, the 356
myofibroblast markers Acta2 and Tagln were expressed by a portion of FB3. These data 357
support the presence of previously described, mutually exclusive, inflammatory (FB1) and 358
myofibroblastic (FB3) CAF subtypes [22-24]. Interestingly, FB3 also expressed numerous MHC-359
II–associated genes (Fig. 5C). GO analysis suggested that FB1 was involved in an acute phase 360
response and inflammatory response, FB2 was more associated with physiological functions of 361
fibroblasts, while FB3 had antigen processing and presentation through the MHC-II pathway 362
All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.. https://doi.org/10.1101/539874doi: bioRxiv preprint
12
and had complement activation functions (Fig. 5D). Furthermore, we analyzed genes that 363
increased in FB1 and FB3 during PDA progression, and found that FB1 showed a progressive 364
increase in the expression of genes associated with inflammatory response and chemotaxis 365
(Fig. 5E and Supplementary Fig. 4B) while FB3 genes displayed increased function on 366
translation during disease progression, possibly due to enhanced antigen processing activity. 367
These data suggest that FB1 is an inflammatory population and the inflammatory feature 368
increases during PDA progression, while FB3 consists of the well-studied myofibroblast 369
population, and displays an enrichment for class 2 MHC genes. 370
371
We also found that some genes essentially exclusive to FB3 in the normal and early KIC 372
pancreas became expressed in FB1 and FB3 populations in late KIC, marking these genes as 373
potential global fibroblast markers in advanced PDA. One such gene was Cdh11 (Fig. 5C). We 374
validated these data by immunohistochemistry. We found in late KIC tumors, stromal staining 375
for αSMA and PDGFRα were nearly mutually exclusive, whereas CDH11 showed uniform 376
staining across all morphologically discernable fibroblasts (Fig. 5F). Taken together, these data 377
provide the first in vivo description of all CAF populations during PDA progression. 378
379
Mesenchymal cancer cells and CAFs show evidence of increased epigenetic regulation and 380
super-enhancer activity in advanced PDA 381
Unique molecular identifiers (UMI) serve to barcode each input mRNA molecule during cDNA 382
library generation, enabling the determination of initial transcript number even after cDNA library 383
amplification [25]. We compared UMI counts across all cell types between early and late KIC 384
tumors (Fig. 6A and B). In early lesions, there was a marked increase in UMI in the beta islet 385
cells (median: 2849, range: 1322-12,857), which might indicate that increased transcriptional 386
activity is a means by which the endocrine requirements of these cells are met. No other cell 387
population in the early KIC tumor displayed this level of UMI. The early KIC cancer cells 388
displayed a relatively low UMI count (median: 1979, range: 1163-7735). In contrast, the 389
mesenchymal cancer cell population in the late KIC tumor displayed a marked increase in total 390
UMI count with a median count of 18,334 and range of 4433-50,061 (Fig. 6C). The epithelial 391
cancer cells in the late KIC also displayed an increased UMI, albeit to a far lesser degree than 392
the mesenchymal cancer cell population (median: 10,368, range: 4940-30,440). 393
394
We reasoned that the increased transcriptional activity may be associated with increased 395
activity of epigenetic regulation as well as super-enhancer [26]. BRD4 belongs to the 396
All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.. https://doi.org/10.1101/539874doi: bioRxiv preprint
13
bromodomain family of transcriptional regulators and is a key regulator of super-enhancer 397
activity [27]. Prior studies have shown that MYC activity is promoted by super-enhancer activity 398
in PDA [28]. We found that in late KIC and KPfC tumors, Brd4 was expressed highly in epithelial 399
and mesenchymal cancer cells while Myc was expressed mainly in the mesenchymal cancer 400
cell population (Fig. 6C, Supplementary Fig. 6). In addition, several genes encoding high-401
mobility group A proteins (Hmga1, Hmga1-rs, Hmga2) were markedly expressed in late KIC and 402
KPfC mesenchymal cancer cells. HMGA proteins are chromatin-associated proteins that 403
regulate transcriptional activity, including enhancesome formation [29]. Lastly, critical 404
components of the SWI/SNF complex (Smarcb1, Arid1a, Arid2), which are essential in 405
nucleosome remodeling and transcriptional regulation [30], were also expressed highly in 406
epithelial and mesenchymal cancer cells of the late KIC, but not cancer cells in the early KIC 407
lesion. Taken together, these data provide multiple lines of evidence to suggest that the 408
transcript load of a more aggressive mesenchymal cancer cell population is increased relative to 409
cancer cells in early lesions or epithelial cancer cells in advanced PDA. 410
411
Interestingly, we also noted that fibroblasts in late KIC tumors also showed an increased UMI 412
(median: 14,538, range: 4461-37,497). They also displayed an increased expression of super-413
enhancer and other epigenetic transcriptional regulator genes in contrast to fibroblasts from 414
normal mouse pancreas or early KIC pancreas (Fig. 6D). These data are suggestive of 415
increased super-enhancer and transcriptional activity as normal pancreas fibroblasts become 416
CAFs. 417
418
We validated these single-cell RNA expression data using three-color immunohistochemical 419
analysis of late KIC tumors: SOX9 was used as a pan-cancer cell marker, vimentin as a 420
mesenchymal marker, and BRD4 was a surrogate marker for super-enhancer activity. We 421
identified positive co-staining for vimentin and Brd4 in CAFs, positive triple-staining 422
(vimentin+/Sox9+/Brd4+) in mesenchymal cancer cells, and single staining of Sox9 in epithelial 423
cancer cells that localized to more differentiated, duct-like structures in the advanced tumors 424
(Fig. 6E). Next, we performed immunohistochemical analysis on 16 whole tumor human 425
pancreatic cancer sections using an antibody against H3K27ac, a commonly accepted marker 426
of increased gene regulatory element activity [26, 31]. The malignant epithelium and stromal 427
fibroblasts were scored separately. These analyses showed markedly positive 3+/3+ staining in 428
the stromal fibroblasts of all whole tumor sections (Fig. 6F). In 6/16 cancer epithelia the score 429
was 1+ and 10/16 scored 2+, with no samples showing a cancer epithelial scoring of 3+. Taken 430
All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.. https://doi.org/10.1101/539874doi: bioRxiv preprint
14
together, these are the first data indicating differential super-enhancer activity in distinct tissue 431
compartments of PDA. 432
433
Discussion 434
We have carried out an scRNA-seq of different stages of the KIC GEMM, in addition to late 435
KPfC and KPC tumors in an effort to agnostically profile the phenotypic changes of cancer and 436
stromal cells during PDA progression. We have established the emergence of a mesenchymal 437
cancer cell population as a late-stage tumor event and have identified novel features of different 438
macrophage and fibroblast populations. This significantly improves our understanding of PDA 439
progression and lays the foundation for the development of novel therapeutic approaches. 440
441
PDA pathogenesis involves metaplasia of normal acinar cells to ductal epithelium, which in turn 442
undergo neoplastic transformation in a KRAS-driven manner [32]. Malignant ductal epithelium 443
may then assume more aggressive, mesenchymal features as the disease progresses. In this 444
study, mesenchymal cancer cell populations were noted in late-stage tumors. Our data support 445
a model in which mesenchymal features of cancer cells are acquired later in the disease 446
process, although others have argued that this can be one of the earliest events in PDA [33]. 447
Mesenchymal cancer cell populations have been studied extensively in pancreatic cancer 448
mouse models and has been shown to be critical to chemotherapeutic resistance while their 449
contribution to metastasis has been more controversial [34, 35]. Mesenchymal cancer cells 450
have previously demonstrated an increased protein anabolism and activation of the 451
endoplasmic reticulum–stress-induced survival pathways in a PDA GEMM [36]. 452
453
Indeed, in the late KIC model, ribosomal pathways were the most significantly upregulated 454
pathways in cancer cells (Fig. 2E-H). It is likely that the demand for increased ribosomal activity 455
stems from high transcriptional activity governed by epigenetic mechanisms in the 456
mesenchymal cancer cells, as we also saw markedly increased UMI counts in this population 457
(Fig. 6B). The bromodomain and extraterminal (BET) family of proteins such as BRD4, which is 458
markedly upregulated in cancer cells and fibroblasts of late-stage PDA (Fig. 6C and D), serve to 459
recruit regulatory complexes to acetylated histones at enhancer sites, resulting in increased 460
transcription [37]. Previously, a combination approach using a BET protein inhibitor and a 461
histone deacetylase inhibitor led to near-complete tumor regression and improved animal 462
survival in a PDA GEMM [28]. Super-enhancer activation has recently been shown to be 463
fundamental in the pathophysiology of a variety of neoplasms [38] and is intimately associated 464
All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.. https://doi.org/10.1101/539874doi: bioRxiv preprint
15
with Hmg2a in PDA, as super-enhancer attenuation has been demonstrated to downregulate 465
Hmg2a expression and the growth of PDA cells in a three-dimensional in vitro model [39]. Our 466
study is the first to demonstrate that these epigenetic regulatory mechanisms in PDA are 467
present in specific tissue compartments, namely mesenchymal cancer cells and CAFs. Future 468
efforts to target super-enhancer activity in PDA should consider distinct tissue compartments 469
governing the sensitivity and resistance to novel therapeutics. 470
471
Our data revealed two molecular subtypes of macrophages in advanced PDA (Fig. 4). One 472
expressed numerous chemokine and inflammation-associated genes while the other was rich in 473
MHC-II–associated genes. In a previous study, MHC-II–positive macrophages were isolated 474
from orthotopic breast tumors and highly expressed CCL17, consistent with our data [40]. In 475
parallel with our study, MHC-IIlow macrophages were found to be highly enriched for numerous 476
chemokines. Moreover, in an orthotopic hepatoma mouse model, an early MHC-II+ macrophage 477
population appeared to suppress tumor growth but an MHC-IIlow macrophage population 478
became the predominant macrophage population as the tumor progressed, resulting in a 479
protumor phenotype [41]. Nonetheless, to confirm their pathophysiological significance, 480
functional studies are required in which inducible selective ablation [42] is performed on the two 481
late-stage PDA macrophage subpopulations using specific markers we have identified in this 482
study. Zhu and colleagues [43] have shown that bone marrow-derived monocytes make up 483
approximately 80% of MHC-II–positive macrophages in a PDA GEMM whereas MHC-II–484
negative macrophages in normal pancreas and PDA were shown to be maintained 485
independently of monocyte contributions. Monocyte-independent MHC-IIlow tissue resident 486
macrophages expanded during tumor progression and contributed to PDA growth and survival. 487
Conversely, Sanford and colleagues [44] have shown that monocytes can give rise to a pro-488
inflammatory macrophage population in a PDA mouse model, which when antagonized with 489
neutralizing antibodies against CCR2, resulted in decreased tumor growth and reduced 490
metastases in vivo [44]. These data highlight the need for an scRNA-seq study on macrophage 491
populations in PDA GEMMs with labelled bone marrow replacement to reconcile these 492
discrepancies. 493
494
More importantly, in the studies of tumor-associated macrophages, inflammatory chemokines 495
are commonly used to indicate an M1 type of macrophage, which are normally associated with 496
immune-stimulatory functions. Nevertheless, our study indicates that a distinct M1/M2 497
macrophage phenotype is not readily discernable at the single-cell level. Instead, as PDA 498
All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.. https://doi.org/10.1101/539874doi: bioRxiv preprint
16
progresses, an inflammatory feature is substantially increased, and this accompanies an 499
increase of an important M2 macrophage marker, ARG1 (Fig. 4F and H). This raises questions 500
on the M1/M2 classification system, as the inflammatory feature is associated with the 501
progression of PDA. Future studies should focus on the function of these inflammatory 502
macrophages in PDA in addition to validating markers for macrophage classification. 503
504
While numerous studies have generally shown that CAFs are tumor-promoting in the biology of 505
PDA and other carcinomas [45, 46], recent studies have found that the function of CAFs in PDA 506
biology are more varied. Özdemir and colleagues [42] demonstrated that the depletion of 507
αSMA+ cells from the microenvironment in a PDA GEMM resulted in shortened survival and 508
poorly differentiated tumors [42], and low myofibroblast tumor content was shown to be 509
associated with worse survival in human PDA sections. These data prompted a paradigm shift 510
whereby certain CAFs may function to constrain, rather than promote, PDA. Moreover, until 511
recently, the molecular heterogeneity of CAFs in PDA has not been well-appreciated. The 512
primary attempt to characterize fibroblast heterogeneity in PDA demonstrated that mouse 513
pancreatic stellate cells (PSCs) could be induced to express αSMA in vitro when directly co-514
cultured with primary mouse PDA cells in an organoid co-culture system [22]. These 515
myofibroblastic CAFs were designated as “myCAFs.” This was distinct from IL6+ fibroblasts that 516
were produced in vitro when PSCs were indirectly co-cultured with mouse PDA organoids 517
through a semi-permeable membrane. The IL6+ fibroblasts were also positive for PDGFRα and 518
numerous other cytokines and therefore termed inflammatory CAFs or “iCAFs.” The 519
immunohistochemistry of human and mouse PDA tissue showed distal IL6+ stroma as a distinct 520
population from the peritumoral αSMA+ stroma. Subsequent studies in PDA GEMMs 521
demonstrated that the iCAF population can mediate pro-tumorigenic properties and is a 522
potential therapeutic target in an attempt to sensitize PDA to immunotherapeutic strategies [23, 523
24]. 524
525
Our current study is the first to demonstrate the existence of three distinct molecular subtypes of 526
fibroblasts in the normal mouse pancreas, which in turn gave rise to two distinct subtypes of 527
CAFs that were largely conserved across three different PDA GEMMs. We noted that FB1 528
expressed insulin-like growth factor signaling genes (Igfbp7, Igfbp4, and Igf1) in addition to 529
Pdgfra, Cxcl12, Il6, and several other cytokines (Ccl11, Ccl7, Ccl2, and Csf1). We propose that 530
our FB1 population is the previously described iCAF population and hence likely pro-531
tumorigenic. Conversely, the FB3 population was positive for the myofibroblast markers Acta2 532
All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.. https://doi.org/10.1101/539874doi: bioRxiv preprint
17
and Tagln, and therefore most closely represents the previously described myCAF population. 533
Importantly, our agnostic approach did not identify any further putative CAF populations and so 534
we support the two-CAF model proposed by Öhlund and colleagues [22]. 535
536
In summary, this report systematically outlines the cellular landscape during the progression of 537
PDA and highlights the cellular heterogeneity in PDA pathogenesis. As such, future targeted 538
therapeutic strategies should be developed with their intended target subpopulation in mind. 539
540
Address requests for reprints to: Rolf A Brekken, 6000 Harry Hines Blvd., Dallas, TX 75390-541
8593. Email: [email protected]. 542
543
Acknowledgements 544
We thank the McDermott Center Next-Generation Sequencing Core at UT Southwestern for 545
preparing and sequencing the single-cell RNA-seq libraries. We thank Dr Jeon Lee (UT 546
Southwestern) for assistance in pre-processing of scRNA-seq data. We also thank Dave Primm 547
of the UT Southwestern Department of Surgery for help in editing this article. 548
549
Conflicts of interest 550
The authors have no conflicts of interest to report. 551
552
Ethics approval 553
Mice were used in this study according to UT Southwestern institutional guidelines and 554
approved by the institutional animal care and use committee at UT Southwestern Medical 555
Center. All human samples were procured through the UT Southwestern tissue management 556
shared resource and approved through the UT Southwestern institutional review board. 557
558
Data sharing statement 559
There are no additional unpublished data from this study. 560
561
References: 562
1. Pancreatic cancer. Nature Reviews Disease Primers, 2016. 2: p. 16023. 563
2. Mei, L., W. Du, and W.W. Ma, Targeting stromal microenvironment in pancreatic ductal 564
adenocarcinoma: controversies and promises. J Gastrointest Oncol, 2016. 7(3): p. 487-94. 565
3. Valkenburg, K.C., A.E. de Groot, and K.J. Pienta, Targeting the tumour stroma to improve cancer 566
therapy. Nat Rev Clin Oncol, 2018. 15(6): p. 366-381. 567
All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.. https://doi.org/10.1101/539874doi: bioRxiv preprint
18
4. Potter, S.S., Single-cell RNA sequencing for the study of development, physiology and disease. 568
Nat Rev Nephrol, 2018. 14(8): p. 479-492. 569
5. Ostapoff, K.T., et al., Neutralizing murine TGFbetaR2 promotes a differentiated tumor cell 570
phenotype and inhibits pancreatic cancer metastasis. Cancer Res, 2014. 74(18): p. 4996-5007. 571
6. Hingorani, S.R., et al., Preinvasive and invasive ductal pancreatic cancer and its early detection in 572
the mouse. Cancer Cell, 2003. 4(6): p. 437-50. 573
7. Hingorani, S.R., et al., Trp53R172H and KrasG12D cooperate to promote chromosomal instability 574
and widely metastatic pancreatic ductal adenocarcinoma in mice. Cancer Cell, 2005. 7(5): p. 469-575
83. 576
8. Butler, A., et al., Integrating single-cell transcriptomic data across different conditions, 577
technologies, and species. Nat Biotechnol, 2018. 36(5): p. 411-420. 578
9. Huang da, W., B.T. Sherman, and R.A. Lempicki, Systematic and integrative analysis of large 579
gene lists using DAVID bioinformatics resources. Nat Protoc, 2009. 4(1): p. 44-57. 580
10. Zheng, G.X., et al., Massively parallel digital transcriptional profiling of single cells. Nat Commun, 581
2017. 8: p. 14049. 582
11. Kopp, J.L., et al., Identification of Sox9-dependent acinar-to-ductal reprogramming as the 583
principal mechanism for initiation of pancreatic ductal adenocarcinoma. Cancer Cell, 2012. 584
22(6): p. 737-50. 585
12. Weniger, M., K.C. Honselmann, and A.S. Liss, The Extracellular Matrix and Pancreatic Cancer: A 586
Complex Relationship. Cancers (Basel), 2018. 10(9). 587
13. Shen, W., et al., TGF-beta in pancreatic cancer initiation and progression: two sides of the same 588
coin. Cell Biosci, 2017. 7: p. 39. 589
14. Gruber, R., et al., YAP1 and TAZ Control Pancreatic Cancer Initiation in Mice by Direct Up-590
regulation of JAK-STAT3 Signaling. Gastroenterology, 2016. 151(3): p. 526-39. 591
15. Gao, J.J., et al., Regulation of gene expression in mouse macrophages stimulated with bacterial 592
CpG-DNA and lipopolysaccharide. J Leukoc Biol, 2002. 72(6): p. 1234-45. 593
16. Leask, A. and D.J. Abraham, All in the CCN family: essential matricellular signaling modulators 594
emerge from the bunker. J Cell Sci, 2006. 119(Pt 23): p. 4803-10. 595
17. Regn, M., et al., Peptidase inhibitor 16 is a membrane-tethered regulator of chemerin processing 596
in the myocardium. J Mol Cell Cardiol, 2016. 99: p. 57-64. 597
18. Li, Y., J. Wang, and K. Asahina, Mesothelial cells give rise to hepatic stellate cells and 598
myofibroblasts via mesothelial-mesenchymal transition in liver injury. Proc Natl Acad Sci U S A, 599
2013. 110(6): p. 2324-9. 600
19. Simpkins, S.A., et al., Clinical and functional significance of loss of caveolin-1 expression in breast 601
cancer-associated fibroblasts. J Pathol, 2012. 227(4): p. 490-8. 602
20. Chang, S.K., et al., Cadherin-11 regulates fibroblast inflammation. Proc Natl Acad Sci U S A, 2011. 603
108(20): p. 8402-7. 604
21. Kanzaki, R., et al., Gas6 derived from cancer-associated fibroblasts promotes migration of Axl-605
expressing lung cancer cells during chemotherapy. Sci Rep, 2017. 7(1): p. 10613. 606
22. Ohlund, D., et al., Distinct populations of inflammatory fibroblasts and myofibroblasts in 607
pancreatic cancer. J Exp Med, 2017. 214(3): p. 579-596. 608
23. Feig, C., et al., Targeting CXCL12 from FAP-expressing carcinoma-associated fibroblasts 609
synergizes with anti-PD-L1 immunotherapy in pancreatic cancer. Proc Natl Acad Sci U S A, 2013. 610
110(50): p. 20212-7. 611
24. Djurec, M., et al., Saa3 is a key mediator of the protumorigenic properties of cancer-associated 612
fibroblasts in pancreatic tumors. Proc Natl Acad Sci U S A, 2018. 115(6): p. E1147-E1156. 613
25. Kivioja, T., et al., Counting absolute numbers of molecules using unique molecular identifiers. Nat 614
Methods, 2011. 9(1): p. 72-4. 615
All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.. https://doi.org/10.1101/539874doi: bioRxiv preprint
19
26. Hnisz, D., et al., Super-enhancers in the control of cell identity and disease. Cell, 2013. 155(4): p. 616
934-47. 617
27. Loven, J., et al., Selective inhibition of tumor oncogenes by disruption of super-enhancers. Cell, 618
2013. 153(2): p. 320-34. 619
28. Mazur, P.K., et al., Combined inhibition of BET family proteins and histone deacetylases as a 620
potential epigenetics-based therapy for pancreatic ductal adenocarcinoma. Nat Med, 2015. 621
21(10): p. 1163-71. 622
29. Ozturk, N., et al., HMGA proteins as modulators of chromatin structure during transcriptional 623
activation. Front Cell Dev Biol, 2014. 2: p. 5. 624
30. Masliah-Planchon, J., et al., SWI/SNF chromatin remodeling and human malignancies. Annu Rev 625
Pathol, 2015. 10: p. 145-71. 626
31. Mack, S.C., et al., Therapeutic targeting of ependymoma as informed by oncogenic enhancer 627
profiling. Nature, 2018. 553(7686): p. 101-105. 628
32. Reichert, M. and A.K. Rustgi, Pancreatic ductal cells in development, regeneration, and 629
neoplasia. J Clin Invest, 2011. 121(12): p. 4572-8. 630
33. Rhim, A.D., et al., EMT and dissemination precede pancreatic tumor formation. Cell, 2012. 148(1-631
2): p. 349-61. 632
34. Izumiya, M., et al., Chemoresistance is associated with cancer stem cell-like properties and 633
epithelial-to-mesenchymal transition in pancreatic cancer cells. Anticancer Res, 2012. 32(9): p. 634
3847-53. 635
35. Zheng, X., et al., Epithelial-to-mesenchymal transition is dispensable for metastasis but induces 636
chemoresistance in pancreatic cancer. Nature, 2015. 527(7579): p. 525-530. 637
36. Genovese, G., et al., Synthetic vulnerabilities of mesenchymal subpopulations in pancreatic 638
cancer. Nature, 2017. 542(7641): p. 362-366. 639
37. Shi, J. and C.R. Vakoc, The mechanisms behind the therapeutic activity of BET bromodomain 640
inhibition. Mol Cell, 2014. 54(5): p. 728-36. 641
38. Sengupta, S. and R.E. George, Super-Enhancer-Driven Transcriptional Dependencies in Cancer. 642
Trends Cancer, 2017. 3(4): p. 269-281. 643
39. Sahai, V., et al., BET bromodomain inhibitors block growth of pancreatic cancer cells in three-644
dimensional collagen. Mol Cancer Ther, 2014. 13(7): p. 1907-17. 645
40. Movahedi, K., et al., Different tumor microenvironments contain functionally distinct subsets of 646
macrophages derived from Ly6C(high) monocytes. Cancer Res, 2010. 70(14): p. 5728-39. 647
41. Wang, B., et al., Transition of tumor-associated macrophages from MHC class II(hi) to MHC class 648
II(low) mediates tumor progression in mice. BMC Immunol, 2011. 12: p. 43. 649
42. Ozdemir, B.C., et al., Depletion of Carcinoma-Associated Fibroblasts and Fibrosis Induces 650
Immunosuppression and Accelerates Pancreas Cancer with Reduced Survival. Cancer Cell, 2015. 651
28(6): p. 831-833. 652
43. Zhu, Y., et al., Tissue-Resident Macrophages in Pancreatic Ductal Adenocarcinoma Originate 653
from Embryonic Hematopoiesis and Promote Tumor Progression. Immunity, 2017. 47(3): p. 597. 654
44. Sanford, D.E., et al., Inflammatory monocyte mobilization decreases patient survival in 655
pancreatic cancer: a role for targeting the CCL2/CCR2 axis. Clin Cancer Res, 2013. 19(13): p. 656
3404-15. 657
45. Olumi, A.F., et al., Carcinoma-associated fibroblasts direct tumor progression of initiated human 658
prostatic epithelium. Cancer Res, 1999. 59(19): p. 5002-11. 659
46. Hwang, R.F., et al., Cancer-associated stromal fibroblasts promote pancreatic tumor progression. 660
Cancer Res, 2008. 68(3): p. 918-26. 661
662
All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.. https://doi.org/10.1101/539874doi: bioRxiv preprint
20
663
All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.. https://doi.org/10.1101/539874doi: bioRxiv preprint
21
Figure Legends 664
665
Figure 1. Cellular heterogeneity during PDA progression. A) Representative H&E sections 666
of the normal pancreas, early KIC lesion, and late KIC lesion (magnification: 20x). B) tSNE plot 667
of normal pancreas displaying 2354 cells comprising 8 distinct cell populations. C) tSNE plot of 668
the early KIC lesion displaying 3524 cells containing 9 cells types with the emergence of the 669
cancer cell population. D) tSNE plot of the late KIC tumor showing 804 cells and 7 distinct 670
populations. Stacked violin plots of representative marker gene expression for each of the cell 671
populations seen in the E) normal pancreas, F) early KIC lesions, and G) late KIC lesions. 672
673
Figure 2. Analysis of early and late KIC cancer cell populations demonstrate the 674
emergence of the mesenchymal cancer cell population as a late event. A) tSNE plots of the 675
early KIC lesion demonstrated the expression of known epithelial markers in the sole cancer 676
population (black outline). Mesenchymal markers were absent in this population. B) tSNE plots 677
demonstrating the emergence of two cancer cell populations in the late KIC tumor. One cancer 678
cell population expressed the epithelial markers (smaller population outlined in black) and a 679
second expressed the mesenchymal markers (larger population outlined in black). C) Violin 680
plots showing the high expression of epithelial markers in the early KIC cancer cell population 681
and late KIC epithelial cancer cell population but not in the mesenchymal population. 682
Mesenchymal markers were overexpressed in the mesenchymal cancer cell population but not 683
in the early KIC or late KIC epithelial cancer cell populations. D) Single-cell profiling heatmap of 684
all early and late KIC cancer cells displaying differentially expressed genes between the three 685
cell populations. Gene names are listed in the boxes on the far right of the heatmap. Each 686
column represents an individual cell and each row is the gene expression value for a single 687
gene. E) KEGG pathway analysis and F) gene ontology analysis comparing all early KIC cancer 688
cells against all late KIC cancer cells. Red bars are increased categories and blue bars are 689
decreased categories. G) KEGG and BIOCARTA pathway analysis and H) gene ontology 690
analysis comparing all early KIC cancer cells against only the late KIC epithelial cells. Red bars 691
are categories increased in the late KIC and blue bars are decreased in the late KIC. (****P < 692
0.0001, ***P < 0.001, **P < 0.01, *P < 0.05). 693
694
Figure 3. Comparison between cancer cells of KIC and KPfC tumors. A) tSNE plot of the 695
late KPfC lesion displaying 2893 cells and 8 distinct cell populations. B) Stacked violin plots 696
showing representative marker gene expression for each of the cell populations seen in the late 697
All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.. https://doi.org/10.1101/539874doi: bioRxiv preprint
22
KPfC lesion. C) Single-gene tSNE plots of the KPfC tumor displaying the presence of epithelial 698
markers (Ocln, Gjb1, and Tjp1) in the epithelial cancer cell population (upper black outlined 699
population) and mesenchymal markers in the mesenchymal cancer cell population (lower black 700
outlined population). D) Violin plots showing the overexpression of epithelial markers in the 701
epithelial cancer cell population and mesenchymal markers in the mesenchymal cancer cell 702
population. E) KEGG and BIOCARTA pathway analysis comparing all late KIC to all late KPfC 703
cancer cells. Red bars are categories increased in the late KIC and blue bars are increased in 704
KPfC. F) Single-cell profiling heatmap comparing all cancer cells in the KIC versus all cancer 705
cells in the KPfC. Each column represents an individual cell and each row is the gene 706
expression value for a single gene. (****P < 0.0001, ***P < 0.001, **P < 0.01, *P < 0.05). 707
708
Figure 4. scRNA-seq analysis of KIC tumor progression reveals multiple subpopulations 709
of macrophages. A) tSNE plot of three macrophage subpopulations in the early KIC tumor. B) 710
Heatmap depicting the 30 top significantly overexpressed genes in each of the three early KIC 711
macrophage subpopulations. Macrophages from the normal pancreas are displayed (far left 712
group). Each column represents an individual cell and each row is the gene expression value for 713
a single gene. C) tSNE plot representation of two macrophage subpopulations in the late KIC. 714
D) Heatmap depicting the 30 top significantly overexpressed genes in each of the two late KIC 715
macrophage subpopulations. Each column represents an individual cell and each row is the 716
gene expression value for a single gene. E) GO analysis of biological processes in the three 717
macrophage subpopulations seen in the early KIC. F) GO analysis of biological processes in the 718
two macrophage subpopulations of the late KIC. G) Violin plots of the expression of 719
inflammatory genes and Arg1 comparing the macrophages in early and late KIC. H) GO 720
analysis of biological processes that are upregulated in the late KIC macrophages relative to 721
early KIC macrophages. (****P < 0.0001, ***P < 0.001, **P < 0.01, *P < 0.05). 722
723
Figure 5. Analysis of fibroblasts during PDA progression reveals multiple molecular 724
subtypes. A) All fibroblasts from the normal pancreas, early KIC tumors, and late KIC tumors 725
were projected onto a single tSNE plot with the FB1, FB2, and FB3 populations distinguished by 726
pink, orange, and brown, respectively (upper left panel). Normal pancreas fibroblasts were 727
highlighted in red (upper right panel), early KIC fibroblasts in green (lower left panel) and late 728
KIC fibroblasts in blue (lower right panel). Normal pancreas and early KIC contained fibroblasts 729
in all three groups whereas the late KIC had only FB1 and FB3. B) Heatmap displaying the top 730
significant genes (cutoff: P < 10-40) for each of the three fibroblast populations. Thirty random 731
All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.. https://doi.org/10.1101/539874doi: bioRxiv preprint
23
cells from each fibroblast population are displayed. All three late-cancer GEMMs (late KIC, 732
KPfC, and KPC) display only FB1 and FB3 fibroblast populations. C) Violin plots demonstrating 733
representative marker genes for each fibroblast subtype: FB1 overexpressed cytokines and 734
Pdgfra. FB3 overexpressed mesothelial markers, myofibroblast markers, MHC-II molecules and 735
Cdh11. D) Gene ontology analysis of the top biological processes in each of the three fibroblast 736
subtypes. E) GO analysis of genes upregulated in late FB1 and FB3 compared to early FB1 and 737
FB3 in KIC, respectively. F) Immunohistochemical analysis of PDGFRα, αSMA, and CDH11 738
stained serially on the same slide. Colors were deconvoluted into a single color layer. PDGFRα 739
and αSMA staining were mutually exclusive, whereas CDH11 was a pan-CAF marker 740
(magnification: 20x). (****P < 0.0001, ***P < 0.001, **P < 0.01, *P < 0.05). 741
742
Figure 6. Analysis of transcriptional activity in different stages of PDA reveals differential 743
epigenetic and super-enhancer activity in distinct tissue compartments. A) tSNE plot of 744
UMI counts in the early and B) late KIC. C) Violin plots of epigenetic regulatory genes in the 745
cancer cell populations of normal pancreas, early KIC, and late KIC. D) Violin plots of epigenetic 746
regulator genes in the normal, early fibroblast, and late fibroblast populations showing their 747
upregulation in CAFs. E) Sequential triple immunohistochemical staining on the same late KIC 748
tumor section for cancer cells (SOX9, pink), mesenchymal cells (vimentin, brown) and super-749
enhancer activity (BRD4, blue). Well-differentiated ductal epithelium stained solely for SOX9 750
(green outline). Mesenchymal cancer cells (blue arrows) and CAFs (brown arrows) both show 751
co-staining with BRD4. F) Immunohistochemical analysis of human PDA whole tissue sections 752
using the H3K27ac antibody. These representative figures from four different human PDAs 753
demonstrate the 3+/3+ staining in the stromal fibroblasts (red arrows) with 1-2+ staining in the 754
cancer epithelium (magnification: 20x). 755
756
757
All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.. https://doi.org/10.1101/539874doi: bioRxiv preprint
24
Supplementary Figure 1. Ultrasound image of early KIC mouse (39 days old), 1 day prior to 758
sacrifice. Two-dimensional measurement of the neoplastic lesion is denoted in teal (1.10 mm x 759
0.82 mm). V = ventral surface, D = dorsal surface, S = spleen. 760
761
Supplementary Figure 2. A) tSNE plot of KPC (6 months) displaying 1007 cells making up 8 762
distinct cell populations as indicated. B) Stacked violin plots of marker genes and UMI for each 763
of the 8 KPC populations. 764
765
Supplementary Figure 3. Heatmap depicting gene expression levels (horizontal) of single cells 766
(vertical) in the macrophage populations of the KPfC GEMMs. Proinflammatory (Macrophage 1) 767
and MHC-II–rich (Macrophage 2) subtypes are noted below the heatmap. 768
769
Supplementary Figure 4. A) The relative proportions of FB1, FB2, and FB3 in the normal 770
mouse pancreas, early KIC lesions, and late KIC lesions. B) Genes increased most significantly 771
in the FB1 and FB3 populations as the normal pancreas progressed to early KIC and then to 772
late KIC. 773
774
Supplementary Figure 5. Violin plots depicting gene expression of Myc and epigenetic 775
regulatory genes in the KPfC epithelial and mesenchymal cancer cell populations. 776
777
778
All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.. https://doi.org/10.1101/539874doi: bioRxiv preprint
Normal Pancreas Early PDA Late PDAH
&E
Fibroblasts-2
10.11%
Fibroblasts-1
6.29%
Islet cells
1.15%
T cells
10.07%
B cells
5.52%
Fibroblasts-3 1.36%
Acinar cells 61.68%
Macrophages
3.82%
Fibroblasts-2
26.65%
Lymphocytes
7.71%
Red blood
cells 7.59%
Cancer cells-1
3.73%
Cancer cells-2
27.86%
Macrophages 42.41%
Fibroblasts-1 4.98%
Fibroblasts-3
5.72%
A
B C Dnormal pancreas 2354 cells early KIC 3524 cells late KIC 804 cells
Islet cells
3.83%
Macrophages
27.70%
Acinar cells
13.22%
Endothelial cells
5.16% Fibroblasts-3 1.02%
Fibroblasts-1 11.01%
Cancer cells 9.90%
Beta cells 1.50%
Fibroblasts-2
26.65%
Amy1
Amy2a2
Pyy
Sst
Ins1
Ins2
Sox9
Krt18
Col1a1
Col1a2
Cdh5
Pecam1
Cd14
Cd52
Sox9
Krt18
Col1a1
Col1a2
Cd14
Adgre1
Cd3d
Cd79b
Hba-a1
Hba-a2
E F GAmy1
Amy2a2
Col1a1
Col1a2
Pyy
Sst
Cd3d
Cd3e
Cd19
Cd79b
Cd14
Adgre1
Figure 1
All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.. https://doi.org/10.1101/539874doi: bioRxiv preprint
KIC Cancer Cells
Fold Enrichment
KIC Cancer Cell Progression
KEGG Pathway Analysis
Top Down-regulated
Top Up-regulated
KIC Cancer Cell Progression
Gene Ontology Analysis
Fold Enrichment
Bio
log
ica
l Pro
ce
ss
Ce
llula
r Co
mp
on
en
t
Top Down-regulated
Top Up-regulated
Cancer cells
(epithelial)
Earl
y K
IC
Cancer cells-1
(epithelial)
Cancer cells-2
(mesenchymal)
Cdh1 Epcam Gjb1 Cdh1 Epcam Gjb1
Cdh2 Cd44 Axl Cdh2 Cd44 AxlLa
te K
IC
A B
C D
E F
Cdh1
Aplp2
Sepp1
Gcg
Wfdc18
Zbtb20
Prox1
Cp
Cldn3
Tm4sf4
Ttr
Dcdc2a
Apoe
Ces1d
Mgst1
Ambp
Spp1
Spink1
Cel
Klk1
Zg16
Reg2
Iapp
Gstm1
Ins1
Ctrl
Rnase1
Reg1
Try5
Cela1
Ctrb1
Clps
Prss2
Cela2a
Cela3b
Sycn
Ins2
Cpb1
Try4
Cpa1
Pnliprp1
Pnlip
Cldn10
Wfdc15b
Tmem176a
Tmem176b
Tmem45a
2210010C04Rik
Late Epithelial Cancer Cells
Early Cancer CellsLate Mesenchymal
Cancer Cells Ggct
Ctgf
Meg3
Serpine1
Inhba
Igfbp3
Npy
Areg
Lgals1
Fxyd5
S100a4
Ctla2a
Sox11
Igfbp4
F2r
Axl
Hmga2
Prkg2
Timp1
Ecm1
Cpe
Emp1
Cd44
Plaur
Pmepa1
Sdc1
Vim
Ldha
Pgk1
Lgals3
Pkm
Anxa1
Nbl1
Anxa2
S100a10
Id1
Plac8
Msln
Rbp1
S100a6
Tspan8
Onecut2
Anxa3
2200002D01Rik
Gsto1
S100a11
Tmsb4x
Rpl29
Tpt1
Rbm3
Basp1
Mcpt2
Cldn2
Ly6e
Epcam
Cldn7
Krtcap3
Cldn4
Wfdc2
Tff2
Krt7
Sfn
Krt20
Ly6a
S100a14
Serpinb1a
Krt19
Pglyrp1
Fxyd3
Muc1
Dmbt1
Gsta4
Lgals4
Tff1
Ly6d
Mmp7
Slpi
2210407C18Rik
Gcnt3
Spint2
Aqp5
Mal2
Anxa5
Ifitm3
Sparc
relative expressionlow high
Fold Enrichment
Cellu
lar C
om
po
ne
nt
Bio
log
ica
l Pro
ce
ss
Early and Late KIC
Epithelial Cancer Cell Comparison
Gene Ontology Analysis
Top Down-regulated
Top Up-regulated
Fold Enrichment
BIO
CA
RT
A
KE
GG
Top Down-regulated
Top Up-regulated
Early and Late KIC
Epithelial Cancer Cell Comparison
Pathway AnalysisGH
Figure 2
Cdh1
Epcam
Cldn3
Cdh2
Vim
S100a4
All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.. https://doi.org/10.1101/539874doi: bioRxiv preprint
Cancer cells-1
13.07%
Cancer cells-2
5.22%
Lymphocytes 2.45%
Macrophages
65.64%Granulocytes
3.94%
Fibroblasts-3
1.76% Red blood cells 3.49%
Fibroblasts-1 4.42%
late KPfC 2893 cells
Top Down-regulated in KIC
Top Up-regulated in KIC
KE
GG
B
IOC
AR
TA
Pathway Comparison between
Cancer Cells in Late KIC and Late KPfC
Fold Enrichment
Gm10260
Gm10709
Rps27rt
Gm8797
Mt1
Mt2Rps18−ps3
Gm8730
Ggct
Nrn1
Igfbp3
Meg3
Igfbp7
Itgb1
Tuba1a
Timp1
Cpe
Hmga2
Lgals1
Vim
Igfbp4
Msn
Fbln2
F2r
Axl
Prkg2
Prkg2
Serpine1
Inhba
Id3
S100a4
Fxyd5
Ctla2a
Mcpt8
Ctgf
Ankrd1
Tnc
Npy Prl2c3 Lgals3 Ldha Pgk1
Cdkn2a Cldn6 Lars2 Mcpt1 Mcpt2
AA467197
Uba52
Wfdc17
Lyz2Ccl6Saa3
Cxcl2Cd74
mt-Co1
Phgr1
Prap1
Gkn1
Tff1
Reg1
Reg3b
Reg3g
Agr2
Gkn3
Gpx2
Ltf
Prom1
Pglyrp1
Fxyd3
Gsta4
Lgals4
Muc1
Car2
Dmbt1
Tff2
Pdzk1ip1
Sftpd
Sepp1Aqp5
Cldn4
Sfn
Cldn3
S100a14
Ly6a
Ly6e
Wfdc2
Epcam
Cldn7
Spint2
Krt7
Krt19
Cdh1
Krtcap3
Mal2
2210407C18Rik
Mmp7
Rbp1
Serpinb1a
Apoe Cp Wfdc18 Ttr Tmem45a
Epithelial Cancer Cells
Mesenchymal
Cancer Cells
Epithelial
Cancer Cells
Mesenchymal
Cancer Cells
KIC KPfC
relative expressionlow high
La
te K
PfC
Cancer cells-1
(epithelial)
Cancer cells-2
(mesenchymal)
Ocln Gjb1
Vim Cd44
Tjp1
Axl
KPfC Cancer Cells
Gjb1
Tjp1
Cldn3
Vim
S100a4
Fbln2
Sox9
Krt18
Col1a1
Col1a2
Cd3d
Cd79
Hba-a1
Hba-a2
Adgre1
Fcgr2b
Itgax
S100a9
A
B
C
D
E
F
Figure 3
All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.. https://doi.org/10.1101/539874doi: bioRxiv preprint
Macrophage1
Macrophage2
Late KIC
Macrophage1
Macrophage2
Macrophage3
Early KIC
Early KIC Macrophage Signatures
Gene Ontology Biological Process Analysis
Ma
cro
ph
ag
e 1
Ma
cro
ph
ag
e 2
Ma
cro
ph
ag
e 3
Fold Enrichment
Late KIC Macrophage Signatures
Gene Ontology Biological Process Analysis
Ma
cro
ph
ag
e 2
Ma
cro
ph
ag
e 1
Fold Enrichment
Figure 4
C
E
F H
Fold Enrichment
Early and Late KIC Macrophages Comparison
Gene Ontology Biological Process Analysis
Top Up-regulated
G
A B
D
low
relative expression
high
Late KIC
Macrophage1
Late KIC
Macrophage2 Anxa1
Emp3
Cd44
Sdc4
Itgam
S100a6
Arg1
Wfdc21
Cxcl3
Ier3
F13a1
Ear10
Osm
Ifitm6
Pf4
Chil3
Lrg1
Ccl6
Ccl9
Wfdc17
Slpi
S100a8
Ccl7
Hbb−bs
Lcn2
Ear1
Ccl2
Ear2
Thbs1
Saa3
C1qb
Aif1
Il23a
H2−Aa
H2−Eb1
H2−Ab1
H2−Dma
H2−DMb1
H2−DMb2
Rgs1
Tspan13
Cst3
C1qa
Acp5
Spp1
Plet1
C1qc
Mgl2
Cxcl16
Mmp12
Mmp13
Cd74
AW112010
Adamdec1
Ccr7
Ifitm1
Ccl17
Tbc1d4
Apoe
Il1r2
Arg1
Il1a
Il1b
Il1r2
Il6
KIC Macrophages
Early KIC
Macrophage3
Early KIC
Macrophage1
Normal Pancreas
Macrophage
Early KIC
Macrophage2
relative expressionlow high
Sod2
H2−DMb1
Il1a
Ctsz
Prdx5
Bcl2a1b
Capg
Ccrl2
Fabp5
Timp1
Rnf149
Il1rn
Trem2
Il1b
Tgm2
Tnfaip2
Ctsd
Cd14
Ctss
Thbs1
Lgals3
Spp1
Lyz2
Slc7a11
Clec4e
Ptgs2
Ear1
Ear2
Lyz1
Fn1
Timp2
Ifi27l2a
Emp1
Glul
Ednrb
Cfh
Hbegf
Pf4
Ltc4s
Pltp
Mrc1
C4b
Wfdc17
Hmox1
Ccl4
Ccl8
Gas6
Fcgrt
Sepp1
Cd163
Lyve1
Cd209g
Cbr2
Ccl2
Ccl12
Folr2
F13a1
Ccl7
Cd209f
Fcna
Etv3
Gm2a
H2afy
Rpl4
Gpr132
Crem
Ccnd2
Vps37b
Jak2
Tnpo3
Tmsb10
Tmem123
Mkrn1
Psmb8
Syngr2
Lsp1
Pkib
Sub1
Cytip
Plbd1
Cst3
Ramp3
Plac8
Tspan13
Il1r2
Tbc1d4
Napsa
Ccr7
Ccl17
H2afz
All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.. https://doi.org/10.1101/539874doi: bioRxiv preprint
FB3
FB3
FB1
FB3
FB2
FB1
FB2
FB1
FB2
FB1
FB3
FB2
Fibroblast Clusters Normal Pancreas
Early KIC Late KIC
FB
1F
B3
Fold Enrichment
Gene Ontology Analysis of KIC Fibroblast
Increasing Signature
Fold Enrichment
PDA Fibroblast Signatures
Gene Ontology Biological Process Analysis
FB
1F
B2
FB
3
Ccl7
Ccl2
Cxcl12
Pdgfra
Il6
Msln
Lrrn4
Upk3b
Cd74
H2-Aa
H2-Dmb1
Acta2
Tagln
Cdh11
Late KIC
Cadherin-11/PDGFRα/αSMA Cadherin-11
PDGFRα αSMA
Ifitm2
Mt1
Serpinh1
Junb
Mgp
Il6
Ccl7
Ccl11
Cxcl2
FB1 FB2 FB3low high
relative expression
Apoe
Thbs1
Bgn
Cygb
Cxcl14
Pcolce
Malat1
Gpx3
Itm2a
Ptn
Sfrp1
Smoc2
Jund
Abca8a
Steap4
Srpx
Fosb
Igf1
Top genes
in the heatmap
Other genes
in the clusters
Jun
Egr1
Gdf10
Gpm6b
Ifitm1
Cxcl1
Cyp1b1
G0s2
Emp1
Pi16
Sema3c
Ly6c1
Timp2
Clec3b
Pcolce2
Efemp1
Creb5
Nov
Cd248
Slco3a1
Rpl35a
Gpc6
Scara3
Serpinb6a
Rnase4
Rpl17
Gstm1
Ptgs1
Lrrn4
Gpm6a
Nkain4
Msln
Lgals7
Krt19
Upk1b
Slpi
Bcam
Myrf
Bst1
Cav1
F11r
Arhgdib
H2-DMb1
Rpp25
Hdc
Plxna4
Cd38
Clu
Cd82
Cmbl
C2
Ndufaf1
Csrp2
Spint2
Stmn2
Cdh11
Cxadr
Cd74
C1rb
C1d
C3
Tpm1
Anxa11
Mdh2
Mdk
Lpcat3
H2-Aa
H2-Ab1
H2-Eb1
H2-DMb1
H2-D1
Cyb5r3
Pkm
Atp9a
Mgat4b
Ltbp3
Ces2g
Hist1h1c
Dkk3
Cbx5
Sfxn3
Nxn
Daglb
FB1
FB2
FB3
BA
D
C
E
F
Figure 5
FB3
FB3
All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.. https://doi.org/10.1101/539874doi: bioRxiv preprint
Lymphocytes
Red blood
cells
Cancer cells-1
(epithelial)
Cancer cells-2
(mesenchymal)
Macrophages
Fibroblasts-1
Fibroblasts-3
Late KIC nUMI
Islet cells
Macrophages
Acinar cells
Endothelial cells Fibroblasts-3
Fibroblasts-1
Cancer cells
Beta cells
Fibroblasts-2
Early KIC nUMI
nUMI
Brd4
Myc
Arid1a
Arid2
Smarcb1
Hmga1
Hmga1-rs
Hmga2
nUMI
Brd4
Arid1a
Smarcb1
A B
C D
E
Figure 6
Late KIC
SO
X9
/Vim
en
tin
/BR
D4
FibroblastsMesenchymal Cancer CellsEpithelial Cancer Cells
F
Hu
ma
n P
DA
H3K
27
ac
All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.. https://doi.org/10.1101/539874doi: bioRxiv preprint
Supplementary Figure 1
S
V
D
All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.. https://doi.org/10.1101/539874doi: bioRxiv preprint
A
B
Supplementary Figure 2
Sox9
Krt18
Col1a1
Col1a2
Cd3d
Cd79a
Peacam
Cdh5
Hba-a1
Cd14
Fcgr2b
Fibroblasts-3
4.37%
Fibroblasts-1
5.26%
Macrophages
64.45%
B lymphocytes
9.33%
Red blood
cells 2.48%
T lymphocytes 9.24%
Cancer cells 2.88%
Endothelial cells 1.99%
late KPC 1007 cells
All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.. https://doi.org/10.1101/539874doi: bioRxiv preprint
Supplementary Figure 3
Macrophage 1 Macrophage 2
KPfC Macrophages
All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.. https://doi.org/10.1101/539874doi: bioRxiv preprint
0%
20%
40%
60%
80%
100%
NormalPancreas
Early KIC Late KIC
FB1
FB2
FB3
FB1
FB2
FB3
FB1
FB3
35.41%
56.94%
7.66%
28.47%
68.89%
2.64%
46.51%
53.49%
KIC Fibroblast Populations
during PDA Progression
Normal
Pancreas
Late KIC
KIC FB1 Increasing Signature
Early KIC
Normal
Pancreas
KIC FB3 Increasing Signature
Late KIC
Early KIC
Supplementary Figure 4
A
B
All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.. https://doi.org/10.1101/539874doi: bioRxiv preprint
Supplementary Figure 5
Myc
Brd4
Hgma1
Hgma2
All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.. https://doi.org/10.1101/539874doi: bioRxiv preprint